Bar sweep analysis

Stimulus overview

The bar sweep stimulus consists of a 4 pixel wide bar that moves across the 30 x 30 pixel stimulus area. The bar is presented in 8 orientations, moving in both the forward and reverse direction for each orientation (16 directions total). Multiple speeds are used depending on the protocol version.

Bar sweep schematic: A vertical bar sweeping across the Protocol 2 stimulus area on the G4 LED arena.

Breakdown of process_bars_p2

function resultant_angle = process_bars_p2(exp_folder, metadata, PROJECT_ROOT)

This function processes the bar sweep responses from Protocol 2. It returns resultant_angle — the preferred direction (in radians) from the slow bar sweeps — which is passed to the flash analysis as an input.

Loading the data

The frame position and voltage data are loaded from the experiment’s Log file:

f_data = Log.ADC.Volts(1, :); % frame data
v_data = Log.ADC.Volts(2, :)*10; % voltage data

This is the data across the entire experiment — all repetitions of the flash stimuli, bar sweeps, and bar flashes. The voltage data is multiplied by 10 because the acquisition hardware divides the voltage by a factor of 10 to fit within the ADC input range. The median voltage across the full recording is also calculated as a baseline reference.

Parsing the bar data

The function parse_bar_data extracts the bar sweep responses from the full recording. It uses the frame position data to identify when bar stimuli were presented.

Step 1 — Find the repetition boundaries.

Using the on_off parameter (whether bright or dark stimuli were used), the parser identifies large drops in frame position that occur at the end of each flash stimulus block. These act as landmarks for locating the bar sweep data within each repetition.

idx = find(diff_f_data == drop_at_end);
idx([1,3,5]) = [];

The remaining indices in idx correspond to the end of the 6 pixel flash blocks — the stimulus immediately before the bar sweeps begin.

MATLAB figure of the frame position data over the entire P2 experiment (blue) with vertical lines indicating the 6 timepoints in the variable ‘idx’ (red).

Since each flash has a 440ms interval and each bar sweep is preceded by a 1000ms gap, there is approximately 1440ms between the last flash and the first bar sweep.

Zoomed in view of the figure above showing only the second timepoint in ‘idx’ that corresponds to the end of the last 6 pixel flash in Rep 1.

The start and end timepoints for all bar stimuli within each repetition are found and stored as rep1_rng, rep2_rng, and rep3_rng.

Zoomed in view of frame position data. (Red) Last frame of the last flash of the 6 pixel flashes from Rep 1. (Cyan) Time point after 440ms gap after the last flash and 1000ms before bar stimulus starts. (Magenta) First frame of the first bar stimulus in Rep 1.

Overview of how the timing of the moving bar stimuli are found using the frame position data. The figure shows the end of the 6 pixel flashes, both bar stimuli and the start of the second repetition of the 4 pixel size flashes.

Range of timepoints for the bar stimuli in rep1 (magenta lines), rep2 (red lines) and rep3 (green lines). The third rep uses the last frame of the experiment as the end of it’s range. It doesn’t matter that this includes the interval time at the end because the actual start and stop times of the bar stimuli are found within these ranges in the next step.

Step 2 — Find start/stop of each individual bar sweep.

Within each repetition range, the function identifies individual bar sweeps by looking for large jumps in frame position (absolute difference > 9). This threshold works because bars first appear around frame 10, so the transition from the background frame (frame 1) to the bar frame produces a jump of > 9.

Frame position over one repetition of all of the moving bar stimuli. Red vertical lines indicate the start and end of each individual flash sweep. These values are stored within ‘idxs_all{1,1}’

Step 3 — Extract voltage data per sweep into a cell array.

The indices from step 2 are used to extract voltage data for each bar sweep. A 900ms context window is added before and after each sweep to capture the baseline. The data is assembled into a [32 x 4] cell array called data:

Rows Content Approximate timepoints per cell
1–16 Slow bar sweeps (16 directions) ~41,100 (0.9s + 2.3s + 0.9s at 10 kHz)
17–32 Fast bar sweeps (16 directions) ~29,100 (0.9s + 1.1s + 0.9s at 10 kHz)

Columns 1–3 contain individual repetitions; column 4 contains the mean across repetitions (after trimming to the shortest repetition length).

Plotting the bar data

1. Circular timeseries with central polar plot — plot_timeseries_polar_bars

This function creates a circular arrangement of 16 timeseries subplots, one per direction. Each subplot shows individual repetitions in light grey and the mean response in blue (dark blue for slow, light blue for fast bars). The 900ms context windows are visible either side of each sweep, with thin vertical black lines marking their boundaries.

A polar plot in the centre of the figure shows the directional tuning. To generate the polar values:

  • The mean response for each direction is trimmed to exclude the 0.9s pre-stimulus interval and the last 0.7s of the post-stimulus interval.
  • The 98th percentile of this trimmed voltage is taken as max_v (peak response).
  • The 2nd percentile of the second half of the trimmed data is taken as min_v.
  • The median voltage across the full recording is subtracted from these values.

The function returns max_v and min_v as [16 x n_speeds] arrays.

2. Polar plot with vector sum arrow — plot_polar_with_arrow

Plots the polar data with an arrow indicating the preferred direction. Calls vector_sum_polar to compute the resultant angle, then add_arrow_to_polarplot to overlay the direction indicator. The arrow magnitude is fixed at 30 (matching the default rlim of [0, 30] which fits most data).

3. Heatmap of peak responses — plot_heatmap_bars

Produces a heatmap of the max_v values per direction and speed.

Calculating metrics from the bar data

The bar responses are further processed to quantify direction selectivity. All metrics are calculated for each speed condition.

1. Reorder by sequential angles — align_data_by_seq_angles

The data cell array is initially ordered by stimulus presentation order (forward/reverse pairs per orientation). This function reorders the rows so they correspond to sequential angles (0, π/8, 2π/8, …, 15π/8).

2. Find preferred direction and align to π/2 — find_PD_and_order_idx

This function computes the preferred direction via vector sum and circularly shifts the data so the PD is always positioned at π/2 (top of polar plot). It also computes:

  • Circular variance (CV): \[CV = 1 - \frac{|\sum r_i \cdot e^{i\theta_i}|}{\sum r_i}\] Range 0–1. CV = 0 means sharp tuning (all response in one direction); CV = 1 means broad/uniform tuning.

  • FWHM (via compute_FWHM): the angular width (in degrees) at which the response drops to half the peak value.

  • Von Mises parameters (via circ_vmpar from the Circular Statistics Toolbox): thetahat (mean direction) and kappa (concentration — higher values = sharper tuning).

3. Compute response metrics — compute_bar_response_metrics

This function calculates three additional metrics from the PD-aligned data:

  • Symmetry ratio: compares response magnitudes at paired angles equidistant from the PD (e.g. PD ± π/8, PD ± 2π/8, …). Formula: sym_ratio = 1 - (Σ|diff_pairs| / Σ_all_responses). Range 0–1, where 1 = perfectly symmetric tuning curve.

  • DSI (vector sum method): \[DSI_{vector} = \frac{|\sum r_i \cdot e^{i\theta_i}|}{\sum r_i}\] Range 0–1. Equivalent to 1 - CV.

  • DSI (PD-ND method): \[DSI_{PD-ND} = \frac{R_{PD} - R_{ND}}{R_{PD} + R_{ND}}\] Where RPD is the response in the preferred direction and RND is the response in the opposite (null) direction. Range -1 to 1.

Saved outputs

All metrics are stored in a structure bar_results and saved as peak_vals_<strain>_<on_off>_<date>_<time>.mat in the results folder. The structure contains:

Field Description
bar_results.slow/fast/vfast.magnitude Normalised vector sum magnitude (0–1)
bar_results.slow/fast/vfast.angle_rad Preferred direction (radians)
bar_results.slow/fast/vfast.fwhm Full-width half-maximum (degrees)
bar_results.slow/fast/vfast.cv Circular variance
bar_results.slow/fast/vfast.thetahat Von Mises mean direction
bar_results.slow/fast/vfast.kappa Von Mises concentration parameter
bar_results.slow/fast/vfast.sym_ratio Symmetry ratio
bar_results.slow/fast/vfast.vector_sum Complex vector sum
bar_results.slow/fast/vfast.DSI_vector Direction selectivity index (vector method)
bar_results.slow/fast/vfast.DSI_pdnd Direction selectivity index (PD-ND method)
bar_results.slow/fast/vfast.max_v_polar Peak response per direction
bar_results.median_voltage Baseline voltage (median across recording)
bar_results.resultant_angle PD from slow bars (function output)

Additional saved variables include data (raw cell array), data_aligned (PD-aligned), ord (reordering indices), and d_slow/d_fast/d_vfast (median-subtracted peak responses per direction, circularly shifted so that the PD is at the π/2 position).

Example polar plot of bar sweep responses in a control RNAi fly to P2 in Dec 2024.

Example polar plot of a T4 cell after Octopamine application showing the responses to the 5 different bar sweep speeds in the P2 version used in Dec 2025.